Marc Hoit

This proposal is for the expansion of the GPU computing capacity for our North Carolina State University���s (NC State) High Performance Computing (HPC) service. Our plan is to install 6 Lenovo SR670 nodes (with 12x H100 and 12x L40 NVIDIA GPUs) so that teaching, classroom use, and research (including research from the science drivers) can expand, relieving the current compute congestion on our 9 GPU cards (2x P100, 4x RTX2080, 2x GTX1080, 1x K20). This proposal includes science drivers that are drawn from a number of domains. The expanded GPU capabilities will allow our researchers��� goals to be feasible; faster simulations, simulate larger systems, use of larger datasets for more accurate Artificial Intelligence (AI) models, and processing Cryo-EM images faster for higher-resolution protein/enzyme structures. As a result, the methods/tools covered in this justification vary widely; from ab-initio methods like Density Functional Theory (DFT) and Molecular Dynamics (MD), to neural network (NN) training for AI models, to Cryo-EM processing for structural biology. Due to an insufficient number of newer GPU nodes, some researchers have opted to pay for cloud services, or build their own clusters. These options are an inefficient use of personnel, power, space and funding dollars, relative to a centralized on-prem HPC with sufficient GPU nodes.

The rapid growth and importance of data science in all sectors of society (business, government, education, and research) threatens to transform the future of jobs. Even more worrisome than this changing mix of jobs is the threat that this change will significantly increase the level of income polarization in our society as people struggle to obtain the skills necessary to participate in this new world of jobs. This income gap will be most pronounced in those segments of our society that do not have ready access to educational or community resources; this is especially the case with rural areas and small towns. Since data science is based on having access to data, geographic proximity is not essential, as people do not need to travel to the data. What is essential, however, is that people have access and use of modern cyberinfrastructure (networks, data, compute, local and virtual communities. This in turn also means that it is quite feasible for people in remote rural areas to participate in data science work and have well paying jobs. And the ability of people to 'telework' has significantly increased over the last decade, especially as people balance quality of life issues with job opportunities. This raises a question and opportunity: might the existing Land Grant College structure be augmented and used to provide some solutions to help address the huge issues associated with workforce shortages in data science as well as broaden the geographic base? Can the Land Grant Universities, along with their Cooperative Extension services, provide educational opportunities in data science and help people who live in rural areas and small towns prepare to become part of the workforce for the 21st Century? The proposed Workshop will explore the interest, issues, and feasibility of leveraging the significant resources and infrastructure of the Land Grant Universities and the data science community to help address the educational, community, and workforce needs of the nation in data science over the next two decades. The Taking Data Science to America's Emerging Workforce will bring together a broad array of people and organizations to address many of the issues and challenges. Lastly, the workshop will identify concrete next steps, follow-on efforts, and develop recommendations on how to establish data science communities of practice across the nation.

The rapid growth and importance of data science in all sectors of society (business, government, education, and research) threatens to transform the future of jobs. Even more worrisome than this changing mix of jobs is the threat that this change will significantly increase the level of income polarization in our society as people struggle to obtain the skills necessary to participate in this new world of jobs. This income gap will be most pronounced in those segments of our society that do not have ready access to educational or community resources; this is especially the case with rural areas and small towns. Since data science is based on having access to data, geographic proximity is not essential, as people do not need to travel to the data. What is essential, however, is that people have access and use of modern cyberinfrastructure (networks, data, compute, local and virtual communities. This in turn also means that it is quite feasible for people in remote rural areas to participate in data science work and have well paying jobs. And the ability of people to 'telework' has significantly increased over the last decade, especially as people balance quality of life issues with job opportunities. This raises a question and opportunity: might the existing Land Grant College structure be augmented and used to provide some solutions to help address the huge issues associated with workforce shortages in data science as well as broaden the geographic base? Can the Land Grant Universities, along with their Cooperative Extension services, provide educational opportunities in data science and help people who live in rural areas and small towns prepare to become part of the workforce for the 21st Century? The proposed Workshop will explore the interest, issues, and feasibility of leveraging the significant resources and infrastructure of the Land Grant Universities and the data science community to help address the educational, community, and workforce needs of the nation in data science over the next two decades. The Taking Data Science to America's Emerging Workforce will bring together a broad array of people and organizations to address many of the issues and challenges. Lastly, the workshop will identify concrete next steps, follow-on efforts, and develop recommendations on how to establish data science communities of practice across the nation.

Networking Infrastructure, equipment and implementation grant

NCSU will enhance the NCBP demonstration system produced in 2011 to provide high availability and an interface for mapping, graphing trends, and listing signals that enable reports via visualization.